py3: use email.generator module instead of email.Generator On py2: >>> import email >>> import email.generator as emailgen >>> email.Generator.Generator is emailgen.Generator True email.Generator is not present on Python 3. Differential Revision: https://phab.mercurial-scm.org/D2454

py3: use pycompat.strkwargs to convert kwargs keys to str Differential Revision: https://phab.mercurial-scm.org/D2452

py3: whitelist test-push-http.t as passing Differential Revision: https://phab.mercurial-scm.org/D2451

util: handle fileno() on Python 3 throwing io.UnsupportedOperation Fortunately, the exception exists on Python 2 so we don't have to do something weirder than this. Differential Revision: https://phab.mercurial-scm.org/D2450

wireproto: use %d to encode an int, not a %s Differential Revision: https://phab.mercurial-scm.org/D2449

httppeer: explicitly catch urlerr.httperror and re-raise On Python 3 it seems urllib.error.HTTPError doesn't set the .args field of the exception to have any contents, which then breaks our socket.error catch. This works around that issue. Differential Revision: https://phab.mercurial-scm.org/D2448

hgweb: pass exception message to builtin Exception ctor as sysstr If we don't do this, the bytes gets repr()ed on Python 3 and we get bogus error strings sent to clients. Ick. Differential Revision: https://phab.mercurial-scm.org/D2447

bundle2: part id is an int, use %d to make it bytes Differential Revision: https://phab.mercurial-scm.org/D2446

bundle2: **strkwargs love on various kwargs constructions Differential Revision: https://phab.mercurial-scm.org/D2445

http: drop custom http client logic Eight and a half years ago, as my starter bug on code.google.com, I investigated a mysterious "broken pipe" error from seemingly random clients[0]. That investigation revealed a tragic story: the Python standard library's httplib was (and remains) barely functional. During large POSTs, if a server responds early with an error (even a permission denied error!) the client only notices that the server closed the connection and everything breaks. Such server behavior is implicitly legal under RFC 2616 (the latest HTTP RFC as of when I was last working on this), and my understanding is that later RFCs have made it explicitly legal to respond early with any status code outside the 2xx range. I embarked, probably foolishly, on a journey to write a new http library with better overall behavior. The http library appears to work well in most cases, but it can get confused in the presence of proxies, and it depends on select(2) which limits its utility if a lot of file descriptors are open. I haven't touched the http library in almost two years, and in the interim the Python community has discovered a better way[1] of writing network code. In theory some day urllib3 will have its own home-grown http library built on h11[2], or we could do that. Either way, it's time to declare our current confusingly-named "http2" client logic and move on. I do hope to revisit this some day: it's still garbage that we can't even respond with a 401 or 403 without reading the entire POST body from the client, but the goalposts on writing a new http client library have moved substantially. We're almost certainly better off just switching to requests and eventually picking up their http fixes than trying to live with something that realistically only we'll ever use. Another approach would be to write an adapter so that Mercurial can use pycurl if it's installed. Neither of those approaches seem like they should be investigated prior to a release of Mercurial that works on Python 3: that's where the mindshare is going to be for any improvements to the state of the http client art. 0: http://web.archive.org/web/20130501031801/http://code.google.com/p/support/issues/detail?id=2716 1: http://sans-io.readthedocs.io/ 2: https://github.com/njsmith/h11 Differential Revision: https://phab.mercurial-scm.org/D2444

statichttprepo: move HTTPRangeHandler from byterange and delete the latter It turns out we've been toting around 472 lines of Python just for this 20-ish line class that teaches urllib how to handle '206 Partial Content'. byterange.py showed up in my \sstr\( Python 3 dragnet, and found itself having overstayed its welcome in our codebase. # no-check-commit because we're moving code that has to have foo_bar naming. Differential Revision: https://phab.mercurial-scm.org/D2443

filemerge: do what the context __bytes__ does, but locally str() here is clearly the wrong thing, and I think the code is clearer when it doesn't just depend on the magic __{str,bytes}__ behavior. I decided to grep around for \sstr\( and see what low-hanging fruit that showed me. This was part of that hunt. That grep pattern still has some things worth exploring. Differential Revision: https://phab.mercurial-scm.org/D2442

py3: convert known-int values to bytes using %d I decided to grep around for \sstr\( and see what low-hanging fruit that showed me. This was part of that hunt. That grep pattern still has some things worth exploring. Differential Revision: https://phab.mercurial-scm.org/D2441

py3: hunt down str(exception) instances and use util.forcebytestr I decided to grep around for \sstr\( and see what low-hanging fruit that showed me. This was part of that hunt. That grep pattern still has some things worth exploring. Differential Revision: https://phab.mercurial-scm.org/D2440

subrepo: use util.forcebytestr() instead of str() on exception Differential Revision: https://phab.mercurial-scm.org/D2437

tests: add missing b prefixes in test-commit.t # skip-blame just some b prefixes Differential Revision: https://phab.mercurial-scm.org/D2436

commitextras: fix on Python 3 by using sysstrs for __dict__ ops I'm dubious of the __dict__ shenanigans in use here, but lack the enthusiasm for figuring out why that was done right now. # skip-blame just some r prefixes Differential Revision: https://phab.mercurial-scm.org/D2435

util: use pycompat.bytestr() instead of str() This fixes at least some environment variable prints for util.system() callers on Python 3. Yay! Differential Revision: https://phab.mercurial-scm.org/D2434

showconfig: allow multiple section.name selectors (issue5797) This seems useful and we can disambiguate the output format solely by the type of the command arguments.

showconfig: use set to filter sections and entry names Before, an entry matching the specified section could be printed twice if the selector wasn't unique. "sections" and "items" are renamed because it's hard to distinguish "sections" from the loop variable "section".

util: factor out shellsplit() function It turned out to be more than the simple posix=True|False flag, so let's introduce a platform function. I also made it py3 ready.

diff: do not split function name if character encoding is unknown Only ASCII characters can be split reliably at any byte positions, so let's just leave long multi-byte sequence long. It's probably less bad than putting an invalid byte sequence into a diff. This doesn't try to split the first ASCII slice from multi-byte sequence because a combining character may follow.

histedit: use repo.revs() instead of repo.set() where revisions are needed Follows up 72da480db4a5. This is just a micro optimization, but looks slightly nicer.

acl: replace bare getpass.getuser() by platform function Follows up dbadf28d4db0. bytestr() shouldn't be applied here because getuser() isn't guaranteed to be all in ASCII. This change means GetUserNameA() is used on Windows, but that's probably better than trying to get the current user name in UNIX way.

context: remove basectx.__int__ (API) basectx is the only type in the repo having __int__ implemented. This magic method can result in unexpected coercion. Furthermore, having it implemented is wrong for some contexts, since rev() may return None in some cases. Previous commits removed known cases in core where contexts are coerced to integers. So let's delete basectx.__int__. This commit is a bit dangerous. While the test suite passes, there are likely still some callers in core that rely on __int__ that don't have test coverage. An alternative would be to issue a deprecation warning and let this bake for a few releases. .. api:: context.basectx no longer implements __int__. Context instances will no longer cast to ints. Consumers should call ``ctx.rev()`` instead. Differential Revision: https://phab.mercurial-scm.org/D2433

patchbomb: resolve revs before evaluating %ld revset Weaning off of basectx.__rev__. Differential Revision: https://phab.mercurial-scm.org/D2432

histedit: resolve revs before evaluating %ld revset We want to stop relying on basectx.__int__. That means we can't use the %ld revset operator with an iterable of contexts. So we expand an iterable of contexts into a list of revs before calling into the revset. Perhaps it would be worthwhile to add a revset format operator that recognizes context instances so we can just pass contexts as revset arguments? Differential Revision: https://phab.mercurial-scm.org/D2431

split: use ctx.rev() instead of %d % ctx And rename a variable for clarity while we're here. This is part of our effort to wean off basectx.__int__. Differential Revision: https://phab.mercurial-scm.org/D2430

commands: use ctx.rev() instead of %d % ctx Weaning off basectx.__int__. Differential Revision: https://phab.mercurial-scm.org/D2429

cmdutil: use ctx.rev() instead of %d % ctx Weaning off basectx.__int__. Differential Revision: https://phab.mercurial-scm.org/D2428